linux-kernel - Re: [PATCH 4/4] video: Remove stack VLA usage

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 9 Mar 2018 00:16:21 -0600
From:   "Gustavo A. R. Silva" <gustavo@...eddedor.com>
To:     Tycho Andersen <tycho@...ho.ws>, Joe Perches <joe@...ches.com>
Cc:     "Tobin C. Harding" <me@...in.cc>,
        Florian Tobias Schandinat <FlorianSchandinat@....de>,
        Bartlomiej Zolnierkiewicz <b.zolnierkie@...sung.com>,
        kernel-hardening@...ts.openwall.com, linux-kernel@...r.kernel.org,
        driverdev-devel@...uxdriverproject.org,
        Kees Cook <keescook@...omium.org>
Subject: Re: [PATCH 4/4] video: Remove stack VLA usage


I sent a patch for this six hours ago:

https://patchwork.kernel.org/patch/10268591/

--
Gustavo

On 03/09/2018 12:11 AM, Tycho Andersen wrote:
> On Thu, Mar 08, 2018 at 10:01:07PM -0800, Joe Perches wrote:
>> On Fri, 2018-03-09 at 16:50 +1100, Tobin C. Harding wrote:
>>> The kernel would like to have all stack VLA usage removed[1].  The
>>> arrays are fixed here (declared with a const variable) but they appear
>>> like VLAs to the compiler.  We can use a pre-processor define to fix the
>>> warning.
>> []
>>> diff --git a/drivers/video/fbdev/via/via_aux_sii164.c b/drivers/video/fbdev/via/via_aux_sii164.c
>> []
>>> @@ -27,6 +27,9 @@
>>>   
>>>   static const char *name = "SiI 164 PanelLink Transmitter";
>>>   
>>> +/* check vendor id and device id */
>>> +const u8 id[] = {0x01, 0x00, 0x06, 0x00};
>>
>> It seems id is now global in multiple places.
>> Perhaps these should be static.
> 
> Does it even need to be global? Why not just get rid of the indirection and use
> ARRAY_SIZE where we mean it? This seems to work for me,
> 
> diff --git a/drivers/video/fbdev/via/via_aux_sii164.c b/drivers/video/fbdev/via/via_aux_sii164.c
> index ca1b35f033b1..87db6c98d680 100644
> --- a/drivers/video/fbdev/via/via_aux_sii164.c
> +++ b/drivers/video/fbdev/via/via_aux_sii164.c
> @@ -35,10 +35,10 @@ static void probe(struct via_aux_bus *bus, u8 addr)
>                  .addr   =       addr,
>                  .name   =       name};
>          /* check vendor id and device id */
> -       const u8 id[] = {0x01, 0x00, 0x06, 0x00}, len = ARRAY_SIZE(id);
> -       u8 tmp[len];
> +       const u8 id[] = {0x01, 0x00, 0x06, 0x00};
> +       u8 tmp[ARRAY_SIZE(id)];
>   
> -       if (!via_aux_read(&drv, 0x00, tmp, len) || memcmp(id, tmp, len))
> +       if (!via_aux_read(&drv, 0x00, tmp, sizeof(tmp)) || memcmp(id, tmp, sizeof(tmp)))
>                  return;
>   
>          printk(KERN_INFO "viafb: Found %s at address 0x%x\n", name, addr);
> 
> Cheers,
> 
> Tycho
>